STRIDE-QA-Mini is a question-answering dataset for autonomous driving, designed to study the spatiotemporal reasoning capabilities of visual language models (VLMs) in autonomous driving scenarios. The dataset contains 103,220 question-answer pairs and 5,539 image samples. The data is derived from real dashcam footage collected in Tokyo (urban, suburban, and highway environments, in various weather conditions).

Dataset structure

• Object-centric Spatial Question Answering (19,895 question-answer pairs): Relationship between two external vehicles

• Egocentric Spatial Question Answering (54,390 question-answer pairs): The relationship between the own vehicle and another vehicle

• Egocentric Spatiotemporal Question Answering (28,935 question-answer pairs): Future distance and direction prediction task